Detecting Plant disease with Computer vision models

Artificial Intelligence (AI) helps farmers reinvent the way they did farming for decades. Computer vision models built on the basis of Artificial Intelligence & machine learning inspired modernization in agriculture sector.

Do plants fall sick? Yes, they regularly fall sick & protecting them from diseases has become a major challenge for farmers.

Why protect rather than prevent?

Plant diseases one of the world’s most common problem. There are various reasons like weather, soil conditions that trigger the fungal infections that cause emergence of diseases on plant. Farmers protect their farms from these diseases to avoid reduction in their yield, this problem has to be addressed at the earliest to avoid spreading to other plants.

Rose leaf and tomatoes with diesease

“They occur regularly and require constant monitoring, although we observe they spread rapidly. By the time we take action on the disease, it spreads all over the farm destroying even healthy plants” says Ramalingam, an agriculturist.

“By the time we come to know about the type of disease and treat them they spread rapidly and we end up investing in hell lot of fungicides to avoid disease. Worst part is different disease require different kinds of fungicides” says Krishna, an agriculturist.

Is AI boon for farmers?

Yes, we can solve this problem with the help of AI & ML driven machines. Developers are focusing on building advanced computer vision models which detect plant/crop diseases. AI & ML are enabling companies to build models which help identifying from diseases & pests to picking fruits (produce). Developers train their models to detect plant diseases at a very early stage and prevent plant diseases.

Annotation tools are used to annotate objects like image, text. Video etc. to train their models. Quality of the model depends on the type of annotation used to train computer vision models.

Annotating with polygon tool

One of the Best tool for annotating objects for agricultural ML models is Polygon tool. Polygon annotation tools supports to annotate irregular shapes, this tool allows to consider precise and target affected areas of defected plants.

Training computer vision models becomes easy by annotating objects using polygon tool. Some of the use cases are:

Training computer vision models using polygon tool to detect rust disease on rose leaf.

Leaf with rust disease before annotation.
Leaf with disease annotated with Polygon tool.
Leaf with rust disease before annotation.
Leaf with disease annotated with Polygon tool.

Trains Models to: Detect the Pustules in Orange, Yellow, Black, Brown, Black or White in color

Identifies: Rust diseases on Plant

By using polygon annotation tool, models can be trained to detect Pustules in Orange, Yellow, Black, Brown, Black or White in color, this enables the model to detect rust disease that’s ruining the plant. Training the stages of the affected leaf makes the model identify in initial phases itself.

The annotation tool enables the models to train small, dark raised spots on the leaf caused due to wet and warm environments. The trained models help farmers in detecting t he disease and protect plant

The most common problem in plants is Aphids. Detecting them now became easy with the help of AI enabled models. Annotating leafs and stems effected with small green or yellow bugs helps the models to detect the disease of the plant.

Landing Page @

Signup @

Login @

Source: Deep Learning on Medium